www.rsc.org/loc Volume 9 | Number 7 | 7 April 2009 | Pages 849–1020

ISSN 1473-0197

Miniaturisation for chemistry, physics, biology, & bioengineering

DoyleLock release lithography

JonesDEP dispensing of nanodroplets

Wu and YetterLiquid monopropellant microthruster

KraljMultilayer devices for gene expression

www.molecularbiosystems.orgRegistered Charity Number 207890

Celebrating 5 years of publishing, Molecular BioSystems publishes cutting-edge research at the interface between chemistry, the –omic sciences and systems biology.

Fast publication and additional online features, added to the high visibility ensured by indexing in MEDLINE, makes Molecular BioSystemsthe perfect place for your research in subject areas including chemical biology, systems biology, proteomics and genomics, cellular processes and metabolism.

Impact Factor: 4.12 (2007 Thomson (ISI) Journal Citation Reports)

Submit your manuscript at www.rsc.org/ReSourCe, or contact the editorial team at MolBioSyst@rsc.org

Celebrating 5 years

Volum

e4

|Num

ber4|2008

Molecular B

ioSystems

Pages269–352

www.molecularbiosystems.org Volume 4 | Number 4 | April 2008 | Pages 269–352

PAPERVladimir N. Uversky et al.Intrinsic disorder in pathogenic andnon-pathogenic microbes: discoveringand analyzing the unfoldomes ofearly-branching eukaryotes

ISSN 1742-206XHIGHLIGHTIvet Bahar et al.Coupling between global dynamicsand signal transduction pathways:a mechanism of allostery forchaperonin GroEL 1742-206X(2008)4:4;1-A

Molecular BioSystems

Indexed in

MEDLINE!

Molecular BioSystems...

Submit your work today!

Indexedin

MEDLINE

Enhanced html articles

via RSC Prospect

High quality

research

5years of

publishing4.12

COMMUNICATION www.rsc.org/loc | Lab on a Chip

Lock release lithography for 3D and composite microparticles†

Ki Wan Bong, Daniel C. Pregibon and Patrick S. Doyle*

Received 8th December 2008, Accepted 4th February 2009

First published as an Advance Article on the web 13th February 2009

DOI: 10.1039/b821930c

We present a method called ‘‘Lock Release Lithography (LRL)’’

that utilizes a combination of channel topography, mask design, and

pressure-induced channel deformation to form and release particles

in a cycled fashion. This technique provides a means for the high-

throughput production of particles with complex 3D morphologies

and composite particles with spatially configurable chemistries. In

this work, we demonstrate a diverse set of functional particles

including those displaying heterogeneous swelling characteristics

and containing functional entities such as nucleic acids, proteins and

beads.

Particles with three-dimensional (3D) morphologies and configurable

chemistries hold great potential for a host of applications in drug

delivery,1 tissue engineering,2–4 optics5 and electromechanics.6 Of

particular interest, patterned particles with precisely positioned

chemistries could provide the building blocks for self-assembled,

dynamic structures with complex functionality.7 Self-assembled 3D

electronic circuits would give one implication of those applications.8

Despite their enormous potential, 3D and composite particles cannot

be efficiently synthesized on a large scale using current

methodologies.9–13

Multiphoton fabrication9 is a well-known method for synthesizing

3D structures as the technique provides unparalleled control of

morphology in all dimensions. In spite of the advantage, this direct

drawing technique is prohibitively time-consuming. For higher

throughput, 3D particles can be generated using a layer-by-layer

process with photo resists.10,11 Unfortunately, these materials are not

ideal for many applications and the chemical patterning is limited to

layered motifs. Three-dimensional particles can alternatively be

generated using the PRINT method,12 where particles are shaped

using a 3D mold. While Janus PRINT particles can be made by

sequentially filling the mold with different materials, this approach

will only yield simple ‘‘striped’’ material patterns on a particle and, to

the best of our knowledge, particles with more than two stripes have

not been synthesized.13

Here, we introduce lock release lithography (LRL), built off of

continuous-flow14 and stop-flow lithography (SFL).15 LRL signifi-

cantly extends the capabilities of SFL to allow for the production of

3D particles which contain intricately patterned chemical regions,

well beyond a striped motif.

Department of Chemical Engineering, Massachusetts Institute ofTechnology, 77 Massachusetts Avenue, Cambridge, MA, 02139, USA.E-mail: pdoyle@mit.edu; Fax: +1 617 258 5042; Tel: +1 617 253 4534

† Electronic supplementary information (ESI) available: Details of thematerials, the microfluidic device, the stop-flow lithography setup, andthe photopolymerization setup are given as supplementary information.See DOI: 10.1039/b821930c

This journal is ª The Royal Society of Chemistry 2009

LRL uses a combination of channel topography, mask features,

and pressure-induced channel deformation. The process consists of

(1) stopping the flow of a UV-sensitive monomer stream through

a microfluidic channel, (2) lithographically printing structures that are

‘‘locked’’ into regions with multi-level channel topography, and (3)

inducing channel deformation via high pressure to release structures

for harvesting.

In the example shown in Fig. 1, we use a channel with positive

relief features in the topography (i.e. post structures protruding from

the channel ceiling) to lock in an array of particles that are formed

by 75 ms of UV exposure through a transparency mask using

a standard fluorescence microscope. Particle morphology is defined

Fig. 1 Process of lock release lithography. First, structures are poly-

merized by shining bursts of UV light through a transparency mask and

a microscope objective. The particles with shapes determined by the mask

and channel topography, are ‘‘locked’’ by relief structures in the channel

topographies. Particles are ‘‘released’’ with channel deflection after

a relatively high pressure (� 5 psi) is applied to initiate flow. Shown is

a differential interference contrast (DIC) image of a collection of 3D

particles with micro-cavity in the channel reservoir. Scale bars are 100 mm.

Lab Chip, 2009, 9, 863–866 | 863

Fig. 2 DIC and scanning electron microscope (SEM) images of 3D

particles. (a),(b) Squares with 10 mm high pillars using a 20 mm high

channel with negative dot patterns on its ceiling and a square mask. (c)–

(e) Squares with 1 mm high line-space patterns using a 30 mm high channel

with negative line-space patterns on its floor and a square mask. (f),(g)

Table-like 3D particles with 1 mm high line-space patterns on the top and

30 mm high supports on the bottom using a 30 mm high channel with

negative line-space patterns on both sides and a circle mask. (h) Micro-

cups with 30 mm deep voids using a 60 mm high channel with positive dot

patterns on its ceiling and a circle mask. (i),(j) Variants using 30 mm high

channels with same kinds of topographies of negative line-space and dots

on their ceiling, but different ring and cross masks. Scale bars are 200 mm

(a,f,j), 100 mm (c,i), 50 mm (b,d,g,h), and 10 mm (e).

by a combination of mask feature shape and channel topography.

Locked into the three-dimensional relief, particles remain immobi-

lized until a relatively high pressure (�5 psi) is applied to the poly-

(dimethylsiloxane) (PDMS) channel to initiate flow and deflect the

channel beyond the point of particle release. Using an automated

valving system shown in the right top of Fig. 1, the flow is then

stopped via pressure release, and the process is repeated, thus

allowing the formation of 3D particles in an automated, semi-

continuous manner. For more details about the automated system,

see ref. 15. The ratio of monomer which is converted into particles

to that which is used to flush out the channel between polymeri-

zation cycles is approximately 1 : 50.

With the need for periodic channel deformation, this process is

well-suited by stop-flow lithography, but incompatible with contin-

uous-flow techniques.14,16 Recently, continuous-flow lithography was

used to fabricate finned structures in ‘‘railed’’ channels for guided self-

assembly.16 However, without release, this process cannot be used to

generate diverse topographies. For example, the cup-shaped particles

in Fig. 1 cannot be released without channel deflection. Several other

advantages are afforded by stop-flow, including improved resolution

and a higher throughput up to 106 �107 particles per hour using

a single microscope.15

The key to LRL is obtaining sufficient channel deformation to

release polymerized structures. For thin PDMS channels, the defor-

mation is determined by classical elasticity theory17 as Dhmax z0.142PW4/Et3 where P is the pressure, W is the width of the channel,

E is the young’s modulus of PDMS, and t is channel thickness. In

agreement with theory, we observed that particles with 20 mm tall

locking structures were released at 5 psi in a 500 mm wide channel,

which had a ceiling PDMS thickness around 200 mm. For 1 mm wide

channels, the equation reveals that 5 psi pressure could provide

enough deflection to release 3D particles with relief features of � 250

mm. To exploit maximum channel deformation and ensure safe

particle release, particles are polymerized near the channel inlet,

channels are kept fairly wide, and the channel region near the outlet

designed to be taller than the particles.

Because LRL is a variant of flow lithography,14–16 the oxygen

lubrication layer near channel surfaces is expected to be a � 1 mm

thick.14 Thus, the achieved particle size in LRL is limited to

micrometer ranges. Particle and feature size resolutions are dictated

by the optical resolution and the ability to replicate features from the

channel wall/ceilings. The typical optical resolution we achieve with

our setup is�1mm. Replicating 3D features on a wall is limited by the

oxygen inhibition layer which is also �1 mm in the current experi-

ments. Smaller inhibition layers, and hence smaller replicated

features, can be achieved by controlling the ambient oxygen

concentration. Topographical channel features used in LRL are at

least a few microns in size, larger than the UV wavelength (�360 nm),

such that optical interference can be disregarded.

For synthesis, we typically use monomers based on poly(ethylene

glycol), as this material is bio-friendly, highly tunable, and can be

functionalized with a variety of biomolecules.18,19 As shown in Fig. 2,

we synthesized a variety of 3D particles with unique mask shapes and

channel topographies. We used negative dot pattern topographies

with square mask features to generate 3D particles with pillars

(Fig. 2a,b). Positive relief topographies can be used to generate dishes

and cups – particles with voids that could potentially be filled with

active components or cells (Fig. 2h). Importantly, LRL can be per-

formed with the same channel and varying masks to give a variety of

864 | Lab Chip, 2009, 9, 863–866

particle morphologies (Fig. 2i,j). As shown in Fig. 2c and 2d, we

could use very fine 3.5 mm line-space patterns (with 7 mm pitch) to

produce particles with precisely defined linear features. These troughs

could be oriented with the particle edges (Fig. 2c,d) or made to be

oblique by simply rotating the mask with respect to the channel.

To demonstrate that both channel floor and ceiling topographies

could be used to dictate morphology, we synthesized table-like

structures with large relief features on one side and a highly resolved 5

mm line-space pattern on the other (Fig. 2f,g). These particles

demonstrate the ability to combine coarse and fine features into

particle topography. Morphologies generated with LRL can be more

complicated depending on the mold used to generate the channel

topographies, and the mask used to polymerize the particles. Molds

generated using standard lithography can be multi-tiered, rounded,

or slanted,20 while virtually any topography can be achieved using

multiphoton fabrication.9 The transparency masks used to generate

particles can have virtually any two-dimensional shape, can be

grayscale21,22 to provide variability in height along particles, and can

be used in conjugation with interference masks to give finely tuned

microporous structures.23 However, in its current inception, LRL is

not suitable for the preparation of interlocking (like chain links)

features or particles with internal hollow structures.

The most attractive feature of LRL is that the release time is

controllable. Because particle release occurs at a critical pressure

(related to deformation), lower pressures can be used to exchange the

monomer without unlocking the particles – this allows the subse-

quent addition of new chemistries. As such, LRL can be used

This journal is ª The Royal Society of Chemistry 2009

Fig. 3 Synthesis of composite particles. (a),(b) A schematic diagram

showing the synthesis of composite particles. Locked structures with

chemistry 1 are covalently linked to chemistry 2 through mask overlap

and UV exposure after fluidic exchange with low pressure. Then, the

composite structures are released by high pressure in both flows. (c),(d)

DIC and fluorescence microscopy images of the particle shown in (a),(b).

Two streams containing PEG-DA and PEG-DA with rhodamine-labeled

monomer were used to respectively present chemistry 1 and chemistry 2.

(e),(f) Fluorescence microscopy images of composite particles with

‘‘autumn trees’’ in a frame. The six locks appear brighter than the rest of

the fluorescent region since they are thicker. (g) Fluorescence microscopy

image of composite particles with ‘‘spring trees’’ in a frame. Scale bars are

100 mm (c,d,f) and 50 mm (e,g).

Fig. 4 Functional particles. (a) Fluorescence and DIC (insert, outlined

for clarity) images of ‘‘Venn diagram’’ particle demonstrating interwoven

(fluorescent monomer, orange) and excluded chemistries (beads, green) in

polymerization overlap region. (b) Fluorescence and DIC (insert, out-

lined for clarity) images of a DNA detector particle with distinct probe

regions. Shown are fluorescent images of a particle after incubation with

target #1 (green) or both targets #1 (green, insert in the top right corner)

and #2 (red, insert in the top right corner). (c)–(e) Fluorescence images of

particles with pH-responsive fins and a cross-shaped rigid support. The

particle keeps its original 2D circle shape in low pH (c), while in an

alkaline pH, the fins bloom to form a 3D flower-like structure (d). (f)

Fluorescence and DIC (insert, outlined for clarity) images of overlapping

zig-zag-shaped particles with encapsulated entities. One strand contains 2

mm green fluorescent beads while the other has 5 nm red fluorescent

streptavidin protein. In all DIC images, particles have been outlined for

clarity. Scale bars are 50 mm (a,b), and 100 mm (c–f).

efficiently to generate composite particles with multiple precisely

positioned chemistries. The process is shown schematically in Fig. 3a.

First, the multi-inlet channel is filled with chemistry #1 and locked

3D structures are polymerized. Then, by adjusting the pressures of

the inlet streams (but keeping them below � 1 psi), a second chem-

istry replaces the first without displacing the locked particle structure.

A unique mask can be used with this chemistry to polymerize distinct

particle features that are covalently linked to the locked particles via

overlap. Finally, a high pressure flow (� 10 psi) is used to release the

composite particles (Fig. 3b).

To demonstrate the synthesis of composite particles, we used two

chemistries, both with poly(ethylene glycol) diacrylate (PEG-DA)

monomer and one with a fluorescent monomer (rhodamine-acrylate,

orange) to easily distinguish the chemistries via fluorescence after

polymerization. We used positive-relief locks to synthesize patterned

particles with a circular center and square exterior (Fig. 3c,d), and

negative-relief locks to make more intricate particles with interior

features and borders (Fig. 3e,f). Both particle types were generated

using two polymerization steps. This approach can be applied to any

number of unique chemistries. Shown in Fig 3g is one example that

applies three chemistries. The particles were synthesized by the same

process using one more PEG-DA monomer stream with 500 nm

fluorescent beads. In this case, the sequence of polymerization is

PEG-DA with fluorescent rhodamine-acrylate, then PEG-DA with

fluorescent beads, and lastly PEG-DA with no fluorescent entities.

Due to lag times associated with fluidic exchange and mask align-

ment, the throughput for composite particles depends on how many

chemistries are applied. As the number of different chemistries

This journal is ª The Royal Society of Chemistry 2009

incorporated into a particle increases, the throughput decreases.

When two chemistries are used, we can generate � 103 composite

particles with dimensions of 50 mm per hour on our current system.

This can be expedited in the future using dynamic mask systems.24

The overlap regions of multifunctional particles can be designed to

provide interwoven chemistries or excluded chemistries, depending

on the pore size of the initial material and the size of entities included

in subsequent monomers. When the pore size is large enough for

molecules to leech in, the result is an interwoven blend of the two

chemistries. However, when entities included in the monomer blend

are larger than the pore size of the existing structure, those entities are

excluded from the overlap region, resulting in a segregation of

chemistries. Examples of interwoven and excluded chemistries are

shown in Fig. 4a. We prepared ‘‘Venn diagram-like’’ particles to

Lab Chip, 2009, 9, 863–866 | 865

investigate the incorporation of chemistries in the overlap regions.

The first chemistry was a pure PEG hydrogel with a pore size

expected to be � 1 nm.25 The second chemistry was PEG with the

addition of rhodamine-acrylate and 50 nm green fluorescent beads.

As can be seen, the fluorescent monomer penetrated the initial gel

structure and was incorporated in the overlap region while the

colloidal entities were excluded.

We exploited the ability to spatially arrange multiple chemistries in

LRL to prepare particles with diverse functionalities. Fig. 4b shows

a DNA detector particle18 with distinct probe regions. The interior

region and four wings contained DNA probe #1, while the other four

wings contained probe #2. The particles were incubated with target

#1 (which was labeled with green fluorescence, FITC) or both targets

(target #2 was labeled with red fluorescence, Cy3) at 10 nM at 37 �C

for 30 min. Fluorescence images confirm that the target oligomers

hybridize only with their complementary probe regions. We also

generated particles with opposing chemistries – specifically swelling

and non-swelling. Swelling chemistries were achieved using PEG/

acrylic acid monomer blends, which are well-known to be responsive

to changes in pH. We made particles with a cross-shaped support and

pH-responsive, fluorescent fins between each arm (Fig. 4c–e). In

acidic conditions (pH� 3), these particles keep their original 2D body

(Fig. 4c), while in neutral to alkaline conditions (pH � 8), the

responsive fins bloom to form a 3D flower-like structure (Fig. 4d,e).

Lastly, we demonstrate the generation of particles with various

encapsulated entities that are organized in complex hierarchies. We

made particles with overlapping zig-zag-shaped chemistries, one of

which was laden with 2 mm fluorescent beads (FITC, green), and the

other with 5 nm fluorescent protein (streptavidin-phycorytherin, red)

(Fig. 4f). This process can be used to encapsulate living cells, stimuli,

or nutrients with precise control over position, which has implications

for applications in tissue engineering. Compared to other hydrogel

particle-based approaches to engineering tissue constructs,2 we can

prepare more precise and intricate building blocks in a scalable and

highly homogeneous manner.

In conclusion, we have demonstrated that LRL can be used to

easily generate a diverse range of 3D and multifunctional composite

particles through the association of channel topography, mask

features, and pressure-induced channel deformation. The computer-

automated method provides high resolution and high-throughput,

similar to that seen in standard stop-flow lithography.15 Also, locked

structures can be built upon to generate complex, composite particles

with a broad range of potential chemistries, interwoven or excluded,

with incorporated entities including nucleic acids, proteins, or cells.

For example, the length scales in LRL are ideally suited to generating

tissue engineering mesoconstructs each containing multiple cell lines

which are precisely positioned within the particle.2,19 In addition, the

juxtaposition of swelling and stiff materials can be exploited to create

particles that can undergo dramatic shape changes. Degradable

polymers can be used to create microparticles which controllably

evolve or fragment over time. We envision that this technology will

provide a simple but powerful means to mass-produce functional

866 | Lab Chip, 2009, 9, 863–866

units for microfluidic operations, filtration systems, and tissue engi-

neering constructs.

Acknowledgements

We gratefully acknowledge the support of Kwanjeong Educational

Foundation, NSF-NIRT Grant No. CTS-0304128, the MIT Desh-

pande Center, the Singapore-MIT Alliance and Royal Society of

Chemistry Journal Grant.

Notes and references

1 D. A. LaVan, T. McGuire and R. Langer, Nat. Biotechnol., 2003, 21,1184–1191.

2 Y. Du, E. Lo, S. Ali and A. Khademhosseini, Proc. Natl. Acad. Sci.USA, 2008, 105, 9522–9527.

3 N. A. Peppas, J. Z. Hilt, A. Khademhosseini and R. Langer, Adv.Mater., 2006, 18, 1345–1360.

4 W. G. Koh, L. J. Itle and M. V. Pishko, Anal. Chem., 2003, 75, 5783–5789.

5 P. Vukusic and J. R. Sambles, Nature, 2003, 424, 852–855.6 C. M. Waits, B. Morgan, M. Kastantin and R. Ghodssi, Sensors andActuators A, 2005, 119, 245–253.

7 S. C. Glotzer and M. J. Solomon, Nat. Mater., 2007, 6, 557–562.8 D. H. Gracias, J. Tien, T. L. Breen, C. Hsu and G. M. Whitesides,Science, 2000, 289, 1170–1172.

9 C. N. LaFratta, J. T. Fourkas, T. Baldacchini and R. A. Farrer,Angew. Chem. Int. Ed., 2007, 46, 6238–6258.

10 S. L. Tao, K. Popat and T. A. Desai, Nat. Protocols, 2006, 1, 3153–3158.

11 C. J. Hernandez and T. G. Mason, J. Phys. Chem. C, 2007, 111, 4477–4480.

12 J. P. Rolland, B. W. Maynor, L. E. Euliss, A. E. Exner,G. M. Denison and J. M. DeSimone, J. Am. Chem. Soc., 2005, 127,10096–10100.

13 S. E. A. Gratton, S. S. Williams, M. E. Napier, P. D. Pohlhaus,Z. Zhou, K. B. Wiles, B. W. Maynor, C. Shen, T. Olafsen,E. T. Samulski and J. M. Desimone, Acc. Chem. Res., 2008, DOI:10.1021/ar8000348.

14 D. Dendukuri, D. C. Pregibon, J. Collins, T. A. Hatton andP. S. Doyle, Nat. Mater., 2006, 5, 365–369.

15 D. Dendukuri, S. S. Gu, D. C. Pregibon, T. A. Hatton andP. S. Doyle, Lab Chip, 2007, 7, 818–828.

16 S. E. Chung, W. Park, S. Shin, S. A. Lee and S. Kwon, Nat. Mater.,2008, 7, 581–587.

17 W. C. Young, in Roark’s Formulas for Stress and Strain (Eds: H. B.Crawford, S. Thomas), McGraw-Hill, New York, 1989, 457.

18 D. C. Pregibon, M. Toner and P. S. Doyle, Science, 2007, 315, 1393–1396.

19 P. Panda, S. Ali, E. Lo, B. G. Chung, T. A. Hatton,A. Khademhosseini and P. S. Doyle, Lab Chip, 2008, 8, 1056–1061.

20 M. A. Unger, H. P. Chou, T. Thorsen, A. Scherer and S. R. Quake,Science, 2000, 288, 113–116.

21 J. Atencia, S. Barnes, J. Douglas, M. Meacham and L. E. Locascio,Lab Chip, 2007, 7, 1567–1573.

22 D. Dendukuri, P. Panda, R. Haghgooie, J. M. Kim, T. A. Hatton andP. S. Doyle, Macromolecules, 2008, 41, 8547–8556.

23 J. H. Jang, D. Dendukuri, T. A. Hatton, E. L. Thomas andP. S. Doyle, Angew. Chem. Int. Ed., 2007, 46, 9027–9031.

24 S. E. Chung, W. Park, H. Park, K. Yu, N. Park and S. Kwon, Appl.Phys. Lett., 2007, 91, 041106.

25 M. B. Mellott, K. Searcy and M. V. Pishko, Biomaterials, 2001, 22,929–941.

This journal is ª The Royal Society of Chemistry 2009